The 7 Places AI Looks Before It Recommends a Human

An AI doesn't pull a name out of thin air. It checks its sources, like a cautious journalist. Here's a map of where it looks, so you know where to be strong, and just as importantly, where you've probably been quietly neglecting the work.

Think of it like a hiring manager doing reference checks

Imagine a hiring manager considering a candidate they've never met. They don't just read the CV the candidate wrote about themselves, that would be naive. They check LinkedIn, they call a couple of references, they see if the person turns up in industry write-ups, they look at whether the candidate's own portfolio actually backs up the claims. An AI answer engine recommending a person to a stranger is doing something remarkably similar, just automated and at enormous scale. It has seven rough places it checks, and no single one of them, not even your own beautifully written bio, is enough on its own. The candidate with a good CV and nothing else loses to the candidate with an average CV and three solid references almost every time, and the same is true here, machines simply run this same instinct at a scale no single hiring manager ever could.

1. Your owned content

Your site, your articles, your published work. This is where the engine confirms what you claim to do and finds quotable depth in your name. Table stakes, and easy to underinvest in. If this is thin or generic, everything downstream has a weak foundation, since the other six places are largely there to corroborate what your own material already says, not to replace it entirely.

2. Third-party press and publications

Coverage in outlets that aren't you. This is the heavy one. When a credible publication quotes or profiles you, the engine treats it as evidence far stronger than your own homepage. A little genuine press goes a long way here. And it doesn't have to be a national outlet, a well-regarded trade publication or regional paper carries real weight too, especially the smaller and more specific your field is, since the machine cares about relevance and credibility more than raw circulation numbers.

3. Respected lists and rankings

"Top X" lists, industry directories, curated roundups in your field. Being named on a list a machine already trusts is a compact, powerful signal, it's a third party publicly sorting you into the "best" bucket. One well-earned mention on a genuinely respected list can outweigh months of solo content, simply because someone else already did the sorting work and put their own reputation behind the judgment, which is a much harder thing to fake convincingly than a paragraph on your own site.

4. Structured data and knowledge bases

Schema on your pages, Wikidata, knowledge-graph entries. These help the engine resolve you as one clear entity and connect your signals. Less glamorous, quietly decisive, because a machine that's sure who you are is a machine willing to name you. Reference sources like Wikipedia and Wikidata carry particular weight here, precisely because they exist to state facts plainly rather than to persuade anyone of anything.

5. Professional and social profiles

Consistent, current profiles on the platforms your field uses. Not for follower counts, but for corroboration and consistency. When your title and bio match everywhere, your entity gets sharper and more trustworthy. A large following with an inconsistent, out-of-date bio actually confuses the picture more than a small, quiet, accurate profile does, which surprises most people the first time they hear it.

6. Peer and expert references

Other practitioners citing your work, linking to you, mentioning you as a source. This is high-quality regard, peers vouching is among the most credible signals there is, and one of the hardest to fake, which is exactly why it counts so heavily when it does show up. It's also one of the slower ones to build honestly, since it depends on other people's genuine opinion of your work rather than anything you can simply publish or purchase.

7. Fresh, citable pages

Recent, well-structured content, yours and others', that engines can cite right now. Recency matters, especially in source-driven engines. A steady drip of current, quotable material keeps you in the running as the conversation moves. This is also the place where public discussion counts more than people expect, which is why forums and Reddit threads shape AI answers more than most people assume, being current and specific often matters as much as being polished, and a lively, honest thread from last month can outrank a beautifully written page from three years ago.

Myth: your website is the most important place to be strong

A lot of people pour their entire effort into their own website, on the reasonable-sounding theory that it's the one place they fully control, so it should matter most. It's actually the opposite, and that's the part people find hardest to accept. Your own website is the one source the engine has the most reason to discount, because everyone's website says nice things about themselves. The places you don't control, press, lists, peer mentions, are the ones doing the heavy lifting precisely because you couldn't have written them yourself. A gorgeous website with zero outside corroboration is a candidate with a great CV and no references, and hiring managers, human or otherwise, have learned to be cautious about exactly that pattern.

A worked example: two owned-content-only experts

Here's a simple, hypothetical case. Expert A has spent two years building a beautiful personal website, dozens of articles, a polished about page, a confident bio. Expert B has a plainer site with half as much content, but has been quoted twice in a trade publication, appears on one respected industry list, and has a former colleague who publicly credits Expert B's work in an interview. Asked to recommend someone in the field, an AI assistant leans toward Expert B, not because the writing is better, but because three of the seven places it checks now agree independently, rather than one place saying everything about itself. This is a made-up illustration, not a formula, but the shape of it repeats constantly across real fields, which is exactly why a competitor sometimes gets recommended over you despite doing objectively less content work, even when that feels deeply unfair from where you're standing.

How structured data quietly ties everything together

It's worth dwelling on place four, structured data and knowledge bases, because it's the least visible and the most misunderstood. This isn't about tricking an algorithm with hidden code, it's closer to a librarian's index card, a compact, unambiguous record that says this name, this profession, this body of work, all belong to the same one person. Without it, an engine sometimes has to guess whether "J. Smith" mentioned in a podcast is the same "J. Smith" who wrote an article three years ago, and when it isn't sure, it plays safe and treats them as two different, weaker signals instead of one strong one. Getting this right is a big part of what we mean when we talk about how AI decides what's actually true about you, since a fractured identity dilutes every other place on this list too, no matter how good the underlying evidence actually is.

How to use this map

Don't try to dominate all seven at once. Audit where you're strong and where you're a ghost. Almost everyone is solid on owned content and profiles, and weak on press, lists, and peer references, the third-party stuff. That gap is your priority list. Fix the places the machine trusts most, and you become the safe, obvious name to recommend, rather than the impressive-looking name it still hesitates over. Once you know your gaps, it helps to have a short list of the exact questions to test yourself against, which is what the prompts buyers type into AI gives you, and to check whether this work is already paying off using the direct test for whether ChatGPT recommends you. Both take only a few minutes and turn this whole map from theory into a concrete to-do list.

What happens when these seven places disagree

Sometimes your owned content says one thing, a two-year-old press mention says another, and your current social profile says a third. This isn't a rare edge case, it's actually common for anyone whose career has moved even slightly, a title change, a new specialty, a rebrand. When the seven places disagree, an engine doesn't average them into a confident blend, it tends to get cautious and hedge, or it picks whichever version happens to be most repeated across the sources it trusts most. That's an uncomfortable realization for a lot of people, because it means an outdated bio sitting quietly on an old conference page can actively work against a more accurate, current one. Cleaning up old, contradictory mentions is unglamorous work, but it often moves the needle faster than publishing something new, since it removes the noise competing with your best, most current evidence, and it usually takes a single afternoon rather than a whole quarter.

A simple monthly audit you can run in twenty minutes

Take the seven places above and give yourself an honest, plain rating for each, strong, thin, or absent. Most people find they're strong on owned content and profiles, the two easiest to control, and thin or absent everywhere else. Pick the single weakest one and set one small, realistic goal for the month, pitching one journalist, reaching out about one list, asking one respected peer for an honest mention if you've genuinely earned it. Repeat next month with whichever place is now weakest. This isn't a sprint, it's closer to watering different parts of a garden in rotation, and after six months the difference between someone who did this and someone who only tended their own website is usually obvious in the answers AI gives about each of them. Keep a simple note of what you tried each month too, since it turns a vague sense of "I've been working on this" into an actual record you can look back on and learn from.